Interval Timing in Aversive Conditioning: Neural Correlates in Amygdala and Related Networks in Rats☆

Abstract In Pavlovian aversive conditioning, the animal not only learns an association between a neutral conditioned stimulus (CS) and an aversive unconditioned stimulus (US), but also that the CS predicts the time of arrival of the US. Recent data suggest that the CS- US interval is learned at the same time as the association and involves the amygdala (Diaz-Mataix et al., 2013a,b). In search of neural correlates of timing, we recorded neuronal oscillations in two amygdala related networks during associative learning. Our first network of interest is an amygdalo-thalamo-hippocampal network. Indeed, the information from the auditory CS and the US converge on cells of the lateral nucleus of the amygdala (LA), an area critical for learning and storing the CS-US association. The auditory CS reaches the LA after being primarily processed in the auditory thalamus, which projects directly to the LA. Among the extensive afferent and efferent connections of the amygdala, interactions with the hippocampus (CA1) are particularly important for the modulation of explicit memory. Moreover, synaptic plasticity in each of these brain regions (LA, auditory thalamus, CA1) occurs after auditory fear conditioning. Our second network of interest is an amygdalo-prefronto-striatal network. The prelimbic cortex has been shown to be involved in fear expression, whereas the basolateral nucleus of the amygdala (BA) is implicated in the acquisition and expression of aversive memories. Lesions of the dorsal striatum disrupt cued fear conditioning ( Ferreira et al., 2003 ). With regard to interval timing, the striatum and the prefrontal cortex have been studied extensively in timing literature. The implication of these structures in the learning of time has been shown in animals and in humans studies (for review see Buhusi and Meck, 2005 , Coull et al., 2011 ). Indeed lesions of the striatum are known to induce deficits in time perception and memorization. Hippocampal lesions also induce deficits in timing behavior in certain peak interval tasks ( Meck et al., 1984 ). More recently, we showed that the amygdala may also be involved in timing the CS-US interval (Diaz- Mataix et al., 2013). While temporal anticipation is fully expressed behaviorally after overtraining (more than 100 trials), the CS- US time interval is learned in as few as one trial ( Davis et al., 1989 ; Diaz-Mataix et al., in press). In an overtraining paradigm, we found neural correlates of timing between the striatum and the basolateral amygdala ( Knippenberg et al., 2012 ). Here we searched for neural correlates of timing the CS-US interval early in conditioning, i.e., after a few trials. We used a modified Pavolvian fear conditioning paradigm in which a 60-s auditory CS+ (7 kHz) predicted the arrival of a US (footshock) at 30 s and not at the end of the CS; this paradigm allows differentiation between the time of arrival of the US and the onset and offset of the CS. Another 60-s tone called CS- (1 kHz) was never associated with the US. We recorded local field potentials (LFP) during trials without shock in behaving rats before and after conditioning. The rats were first habituated to the two CSs and neural activity was recorded to measure a baseline activity to the CSs alone before conditioning. Then, the rats were first conditioned with the US arriving at 30 s then at 10 s. The rats were tested in extinction trials always 24 hours after conditioning. Frequency analysis was performed on theta (3-9 Hz) and gamma (60-90 Hz) bands, because neural correlates of timing were observed in those bands in overtrained animals ( Knippenberg et al., 2012 ). Power spectrum density analysis showed onset activity in the theta range for LA and thalamus as well as in the gamma range for the thalamus and CA1. A sustained activity was observed in the gamma frequency range for the amygdala. We also analyzed the coherence in oscillatory activity in the LFPs between structures. We found onset responses in the theta range for all pairs of structures. Time locked activity was seen in the theta frequency between the thalamus and the CA1, whereas a sustained activity was seen for the same frequency range between striatum and amygdala. Further analyses are currently being done to refine our results and look at causality and other frequency ranges. At present, our results suggest an involvement of both circuits in processing the CS-US interval early in training during Pavlovian aversive conditioning.